5.8GHz LNA stabilty simulation question

[John_li]

Hi all,
5.8GHz LNA simulation-----if i use s-parameter model(add all match circuits and Bias without DC), 1.321GHz unstable.But if use nolinear model(add all match circuits and Bias),i found it is stable in all frequencies.My question is which one is correct ?
Thank you in advanced,

 

[BigBoss]

S-parameters are genrally measured ones and I think you should trust them.Models sometimes may be inaccurate..

 

[John_li]

S-parameters are genrally measured ones and I think you should trust them.Models sometimes may be inaccurate..

Thank BigBoss for kindness.Today I measured the LNA part of module in MP stage,it is stable in 100KHz~10GHz the same match network and DC bias.It maks me confused which one correct.I use ADS momentum tool.
And Anyone use the same tool simulate correctly compared with product?

 

[hosein_rabiee_g]

I agree with BigBoss. the nonlinear models may be inaccurate.
but you must notice that any unstable network, should not oscillate. and so may be not oscillate when you measured it. a unstable network amplify the noise and the signal at its input and output be large. at large signal state, the s-parameters changed and may be the network become stable. it is damping oscillation.

 

[John_li]

Then how to distinguish unstable case and oscillate?

 

[BigBoss]

Than how to adjust the case of unstability and oscillate?

Look at input and output stability circles and push these circles by connecting series or parallel resistors to outside of smith chart.A small series feedback or a proper parallel feedback helps to block the oscillations too.

 

[John_li]

Thank BigBoss and hosein_rabiee_g for kindness.My confusion is the measurement to judge stability.In fact,I used one product that in MP stage to simulate,normally it is stable in all frequencies,and in fact i used SA to measure it the input connected 50R,output connected to SA,then power on,no spurious component found .But when i used Momentum now.I found 1.3G unstable.Attached my simulation and result.which one is correct,simulation result or measured result(product in MP)?

 

[BigBoss]

Thank BigBoss and hosein_rabiee_g for kindness.My confusion is the measurement to judge stability.In fact,I used one product that in MP stage to simulate,normally it is stable in all frequencies,and in fact i used SA to measure it the input connected 50R,output connected to SA,then power on,no spurious component found .But when i used Momentum now.I found 1.3G unstable.Attached my simulation and result.which one is correct,simulation result or measured result(product in MP)?

The stability can be measured-eventough not practical- up to fmax that is maximum oscillation frequency of active device.So, not to see spurious or anomaly is not a figure of stability concern.
In the other hand, simulations can have some discrepancies and they are just simulations.There may be bad modelled components,simulation set-up errors etc.
As I see, unstability is coming from input side and TL which is connected series to emitter can create unstability, check this one out.The better method to investigate input and output stability circles.
You can connect the output to input to form a closed loop and a AC simulation will give you gain and phase information ( Bode curves) and phse and gain margin to prevent this circuit from oscillation.

 

[John_li]

Thank you for kindness.That means it is hard to predict stability when designing MW amplifier with simulation?IF use ideally schematic to simulate,normally get stability.but use momentum to do it,some frequencies unstable,even though push these circles by connecting series or parallel resistors to outside of smith chart in schematic simulation,so that i am not sure whether
fix or not.And it effects the next step design,

 

[tony_lth]

Momentum simulation is very presice. I believe so, because I have designed X band LNA with it and confirmed that.

 

[John_li]

Momentum simulation is very presice. I believe so, because I have designed X band LNA with it and confirmed that.

Hi tony,
What is your simulation tool?ADS?Have you ever met the same case as me that some frequencies unstable in momentum,but schematic ok?

 

[tony_lth]

I checked the stability with both ADS and AWR.
Momentum also needs to be checked in SCH, doesn't it?
I think the steps should be: Pure SCH --> Momentum SCH -->Final circuit

 

[John_li]

Thank you for kindness.Can you help me find the root cause of instability in 1.3GHz?

 

[rfmw]

Your Stabfact simulation also indicates instability in the whole range from 3GHz to 4GHz, beside the 1.3GHz. You should reconsider the schematic topology. Your base inductance 3nH+3pF gives a resonance at 1.6GHz, including also the input capacitance, that gives you a resonance and instability at roughly 1.3GHz.

 

[John_li]

Thanks for rfmw's help.Which topology i can select that can fixed 1.3GHz and 3~4GHz instability?The collector has been connected 0R .

 

[rfmw]

Look at the LNA application notes from Avago, Infineon, etc.

 

[John_li]

I have found some documents with your suggestion.But the solutions seems not to fix my case.for the collect is connected by 0 R,but still instability in all frequencies.How can i fix the case?

 

[tony_lth]

There are many ways to solve the instable issue, such as replace 0R with 5ohms, or even larger etc.